Database-Recorded BALF Labels and Prolonged Hospital Stay in Pediatric LRTIs

Objective: To evaluate associations between database-recorded bronchoalveolar lavage fluid (BALF) microbiology labels and prolonged hospitalization among children with lower respiratory tract infections (LRTIs) who underwent clinically indicated bronchoalveolar lavage (BAL). Methods: This single-center retrospective data-reconstruction study screened 3,480 hospitalization episodes from March 2023 to March 2025. The cohort was reconstructed from original deidentified hospital and microbiology exports, mixed-flora records were retained, and evaluable protocol eligibility criteria were re-executed. Hospital and microbiology records were matched to the indexed admission and BALF specimen using patient and encounter identifiers, BAL and specimen dates, and specimen type. The final analysis included 1,695 episodes from 1,628 children. Specific microbiology labels could coexist. The primary outcome was hospital stay >14 days. Logistic generalized estimating equations (GEE) clustered by child were used; an extended attenuation model and a length-of-stay sensitivity analysis were also performed. Results: Prolonged hospitalization occurred in 262 episodes (15.5%); median length of stay was 9 days (interquartile range, 7–12). Mixed-flora-only records were present in 1,281 episodes, and 414 had at least one specific label. In the primary GEE model, infancy, clinician-recorded severe pneumonia, major underlying conditions, malnutrition, and several microbiology labels were associated with prolonged stay. After complication, ventilatory-support, and pediatric intensive care unit proxies were added, several estimates attenuated, including Acinetobacter baumannii from 6.35 to 3.61 and Haemophilus influenzae from 1.66 to 1.18. Conclusion: Database-recorded BALF microbiology labels and administrative severity proxies were associated with prolonged hospitalization. Because recorded labels were not adjudicated as causative pathogens, prior antibiotic exposure was unavailable, and assay methods and quantitative interpretation criteria were heterogeneous, the findings are exploratory and do not support pathogen-specific or causal interpretation.